›

Mosaic by Coordinates on JWST data

Derek Warltier
in PixInsight
I am trying to process some JWST data of NGC 1514 and having trouble with Mosaic by Coordinates.  Firstly, I need to state that I cannot plate solve these images because they have too few stars in them,  The PixInsight Plate Solver cannot find enough stars and I have tried using the SetiAstro Blind Solver script, which claims to  be able to solve the image, but cannot generate the calibration data.  I have tried submitting these directly to Astrometry.net, which also fails to generate the calibration data.  I have come across issues like this before where some scripts will not accept such long focal lengths, especially for JWST and HST.  Undaunted, I have trimmed some of the images using a simple Pixelmath expression, which leaves the image essentially untouched except for making some areas completely black.  This action does not affect the astrometric solution.  However, when I try to run some of these images through Mosaic by Co-ordinates, it clearly does not generate the correct mosaic files.  It is my understanding that I should be able to take all of the mosaic files and lay them on top of each other and if the original coordinates were correct, they should be all be lined up, with large black areas.  However, that is not the result I get.  The only hint I have is that when I load these initial images into PixInsight, the console indicates that the images are rotated 0°, but flipped.  Given that I cannot plate solve these, I cannot flip them in the rotate process to see if that helps Mosaic by Coordinates generate the correct size of final image with the frames in the right place.

Any suggestions?

BTW, I have tried Align by Coordinates and that aligns the images correctly (but truncated), but obviously does not generate the full-sized mosaic.

Comments

  • BlockHeadBlockHead
    edited April 30

    JWST comes with a solution already in the data. 
    NGC 1514 is in multiple images? 

    So JWST does not specify the coordinate origins (or flip/mirror) as part of the solutions..they need to be the same. MbC is the correct tool if the coordinate systems match (as is AlignbyCoordinates). 
    I am not certain if MbC works with large overlap.

    I do not understand why you cannot increase the "canvas size" of your images and then use AbC.

    -the Blockhead
  • Derek Warltier
    It's a ten frame by three filter mosaic.  I was thinking of using the crop process to set up the full-sized canvas and place the images in the correct spot (using Excel to calculate the off-sets).  I should be able to use the GradientMergeMosiac at that point.  As far as I can tell, all the images have zero rotation, or nothing BlurX can't fix.  I will give AbC a try as long as increasing the canvas doesn't wipe out the astrometric solution.

    Thanks for your help.
  • BlockHeadBlockHead
    Yeah..it might.
  • BlockHeadBlockHead
    I just don't understand how NGC 1514 (a PNe) can be a 10 frame mosaic...
    NGC 1514 is 2+ arcminutes.
    JWST FOV is 3 arcminutes.


  • Derek Warltier
    Although the MIRI FoV is nearly 2 arcminutes, nearly half of it is unusable. 


    The ten panel mosaic is required to make sure there is usable data for the entire image.  Admittedly, there is quite a bit of overlap, which, I am hoping will help to generate a good image.  I tried uploading a couple of files that show the chip coverage and the mosaic, but couldn't get them to upload.

    You might want to try processing this data yourself.  There are some interesting versions around, including a recent NASA APOD.
  • BlockHeadBlockHead
    What you are describing is what I might call a dither- albeit large.

    Well.. we know that AlignByCoordinates works. 
    Really all you need/want is the working mode of AlignByCoordinates to be Union (Mosaic).
    This is available as  you know in StarAlignment. It seems like a trivial feature request for PI developers.

    -the Blockhead
Sign In or Register to comment.